The NY Times recently covered promising developments in the area of tools that help analyze online news and information.
The article Hot Story to Has Been described several efforts under way that aim to chart the evolution and propagation of stories. I will be writing in more detail about the Web-based tools listed that are publicly accessible in a future post.
The article Mining the Web for Feelings reported on advances in the ability to automatically detect the tone of a story. These tools are variously called sentiment analysis or opinion mining because they go beyond mere facts to infer the tone of articles.
This has been kind of a holy grail for online monitoring and artificial intelligence. It requires analysis of words and context to detect the nuances of sarcasm, for example, and “understand” the sentiment of the writer.
I visited the websites (where available) for each service and plan to eventually kick their tires, write about them, follow their progress and share this info.
The one that seemed most accessible and thus easy to check out and review was Newssift, from the Financial Times Group. The information on the site did not make clear how many sources it searches across. But my quick look and a few searches convinced me that this is a tool worth spending some time with and checking out further.
E.g., plugging in a generic tech term like “cloud computing” revealed overwhelmingly positive sentiments about the topic (see below). Along the left panel it shows the frequency of the term by source type, and shows the top people, companies, themes and regions associated with the term. You can refine the search by selecting specific people, companies, topics etc. that are suggested and displayed along the top panel.
If you select a company, you can see a chart of sentiment trends, making this a great tool for exploring the evolution of brand perceptions. Very cool indeed!
